Machine learning may be the right tool for pr

0


October 28, 2021 – Prescription opioid laws alone cannot predict opioid distribution patterns, according to a new study from the Columbia University Mailman School of Public Health. An analysis of the laws, however, was able to identify access to patient data and several clinical pain management conditions to predict the county’s prescription opioid distribution patterns. Columbia’s research used machine learning approaches (i.e. a series of computer algorithms) to accurately predict patterns and is the first study to rely on machine learning to analyze opioid laws. The results are published in the journal Epidemiology.

Over the past 15 years, in response to the opioid crisis in the United States, states and the federal government have implemented numerous laws regulating the prescribing and dispensing of opioids. These include prescription drug monitoring programs (PDMPs), pain management clinic laws, and limits on initial opioid prescriptions, among others.

“The aim of our study was to identify the combinations of individual and prescription opioid-related legal provisions that were the most predictive of high opioid distribution and high-dose opioid distribution in US counties. said Silvia Martins, MD, PhD, associate professor of epidemiology at Columbia Mailman School. “Our results showed that not all laws on prescription drug monitoring programs are created equally or influence effectiveness, and there is a critical need for better evidence on how variations of the law may affect opioid-related results. We found that a machine learning approach could help identify what determines a successful prescription opioid distribution model. “

Using 162 Prescription Opioid Act provisions capturing prescription drug monitoring program access, reporting and administration functions, pain management clinic provisions and limitations of Prescription opioids, researchers looked at various approaches and models in an attempt to identify the most predictive laws of county level and high doses. dispensation in different epidemic overdose phases – the prescription opioid phase (2006-2009), the heroin phase (2010-2012) and the fentanyl phase (2013-2016) – to further explore the pattern changes in the over time.

The PDMP patient data access arrangements most consistently predicted high-dose and high-dose dispensing counties. The pain management clinic arrangements generally did not predict dispensing metrics during the opioid prescribing phase, but became more discriminatory between high dispensing and high dose counties over time, particularly during the fentanyl period. The predictive performance of the models was poor, suggesting that prescription opioid laws alone do not strongly predict dispensing.

“While further research using various study designs is needed to better understand how opioid laws in general, and in particular, can limit the inappropriate prescribing and dispensing of opioids in order to reduce opioid-related harms,” We are confident that the results of our machine learning approach to identifying salient laws provisions and combinations associated with execution rates will be critical in testing which legal provisions and which combinations of legal provisions work best in future research ” , noted Martins.

Researchers observe that there are at least two major challenges to assessing the impacts of prescription opioid laws on the distribution of opioids. First, states in the United States often pass very different versions of the same general type of law, making it particularly important to examine the specific provisions that make those laws more or less effective in dealing with opioid-related harms. Second, states tend to pass more than one type of law simultaneously, making it difficult to isolate the effect of a specific law or provisions.

Machine learning methods are increasingly applied to similar large-scale data problems and may offer a complementary approach to other forms of policy analysis, notably as a screening tool to identify policies. and the interactions between legislative provisions that require special attention, ”Martins said.

###

Co-authors are Emilie Bruzelius, Jeanette Stingone, Hanane Akbarnejad, Christine Mauro, Megan Marzial, Kara Rudolph, Katherine Keyes and Deborah Hasin, Columbia University Mailman School; Katherine Wheeler-Martin and Magdalena Cerdá, NYU Grossman School of Medicine; Stephen Crystal and Hillary Samples, Rutgers University; and Corey Davis, Public Health Law Network.

The study was supported by the National Institute on Drug Abuse, grants R01DA048572, 1R01DA047347, 1R01DA048860, and K01DA049950; the Agency for Health Care Quality and Research, grant R18 HS023258; and the National Center for Advancing Translational Sciences and the New Jersey Health Foundation, grant UL1TR003017.

Columbia University Mailman School of Public Health

Founded in 1922, Columbia University Mailman School of Public Health pursues a program of research, education, and service to address critical and complex public health issues affecting New Yorkers, the nation, and the world. Columbia Mailman School is the seventh largest recipient of NIH grants among schools of public health. Its nearly 300 multidisciplinary faculty members work in more than 100 countries around the world, addressing issues such as infectious and chronic disease prevention, environmental health, maternal and child health, health policy, change climate and health, and public health preparedness. It is a leader in public health education with more than 1,300 graduate students from 55 countries pursuing a variety of masters and doctoral programs. Columbia Mailman School is also home to many world-renowned research centers, including ICAP and the Center for Infection and Immunity. For more information, please visit www.mailman.columbia.edu.


Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of any press releases posted on EurekAlert! by contributing institutions or for the use of any information via the EurekAlert system.


Share.

About Author

Leave A Reply